Firms in developing countries exhibit large variations in revenue and output, (given the level of their inputs). A number of researchers have suggested that this is due to “capital misallocation”, i.e. capital does not flow to the households, firms, and entrepreneurs who can use it most productively. There are, however, other plausible explanations of the variability of input-adjusted revenue. One is that firms with greater market power (perhaps because they are more innovative) may charge higher prices and thus generate greater revenues from identical production functions. Other explanations centre around firms’ costs of adjusting their capital stocks – if the demand for a firm’s products is highly variable while changes in capital are expensive, firms will choose to maintain apparently inefficient levels of capital during high or low demand periods because they expect demand to return to more normal levels in the future. Another explanation might simply be error in the estimation of the production function – either through measurement errors or model misspecification.
In this pilot project carry out an in-depth quantitative study of a single industry – brick-making in India – that offers the ideal characteristics to measure firm productivity and industry structure far more accurately than has been previously achieved. The results will provide, for the first time, the level of detail necessary to disentangle these competing explanations of the variance of revenue and output across firms. The brick industry has several advantages for this project:
- Bricks are a relatively homogeneous good, and quality differences in the physical product are easily quantifiable.
- Bricks are expensive to transport, making markets relatively segmented. Thus individual towns can be treated as separate markets
- Bricks in India are manufactured at a large range of firm sizes. Thus issues such as inefficient scale, and informal versus formal manufacturing that often feature in debates on the industrial sector in developing nations are directly observable in the brick industry.
- The production function for bricks is relatively simple – mud, fuel, and labour are the three primary inputs – and therefore production functions can be estimated to a degree of accuracy that would be impossible in more sophisticated industries. This is crucial in reducing measurement error.
- Investment in brick manufacturing is very lumpy, with the key production technology, the kiln, occurring in large, discrete and easily measurable units. This gives rise to the dynamic savings and investment decisions that are at the base of many models of productivity dispersion.
We will first carry out a market level census of brick manufacturers. This census will provide market-level information on the number, size, and concentration of firms supplying bricks. Using the census results, a sample of about 600 firms will be selected, stratified on market location and firm size and will be asked to respond to a detailed survey questionnaire, including questions on the characteristics of the firm owner; sources of credit and financing; and pricing decisions, in particular whether these respond to demand shocks (as a measure of market power). The survey will be used to accurately estimate firm productivity.
The unique dataset generated by these surveys would be one of the major outputs of the pilot study. The data collected can provide the basis for a wide range of future papers, focusing not only on productivity but on a wide range of firm level outcomes, much as the ICRISAT data has done for farming. All data and survey formats generated from this project would be posted online to be used by other researchers.
